Japan sold JPY 700bln of 20-year paper this morning and the bid-to-cover came in at 5.84 times, against 4.01 at the last auction and a 12-month average of 3.73. That is a record, and two readers call it one. The part worth your attention is that this record demand arrived at an average yield of 3.7230 per cent — 13.3 basis points below the previous auction. The story being told about the result is that a yield spike lured buyers in. The clearing level says the opposite happened. Both are true, about different quantities, and if you take the narrative and infer the auction cleared cheap you have the sign backwards by 13 basis points.
Three numbers, all moving the same way
The result, as published: JPY 700bln sold; bid-to-cover 5.84 against 4.01 previously; average yield 3.7230 per cent against 3.8560 per cent; tail 0.07 against 0.15.
Work the deltas. Cover is up 1.83 turns, 45.6 per cent better than the September sale and 56.6 per cent above the twelve-month average. The average yield is 13.3 basis points lower. The tail has more than halved, down 53.3 per cent. On every published measure this auction went better than the last one, which is rarer than it sounds — strong cover and a tight tail usually come with a concession somewhere, and there is no concession here.
Two readers carry the cover figures and the tail and they agree to the second decimal. Neither of them is the Ministry of Finance.
The demand story and the clearing level point opposite ways
The attribution offered is that investors stepped in because yields had reached multi-decade highs: the 20-year reached its highest in about thirty years earlier this month and the 30-year touched a record, on inflation worry out of the Middle East and higher crude, plus fiscal worry about the Prime Minister’s plan to cut the consumption tax on food. One insurer is quoted saying higher yields give it room to trim overseas bonds and move into yen debt.
That is a coherent account of the level between auctions, and nothing here disputes it. But it is not an account of this auction, because this auction cleared 13.3 basis points richer than the one it followed. The yield that lured the buyers is the one that prevailed in the weeks before the sale, not the one the sale printed. Those are different measurements of different things, and the sentence “record demand because yields spiked” silently merges them.
The practical version: if your rule is to expect a cheapening into a well-bid long-end auction, this result does not support it, and a rule keyed to “demand was strong so the concession must have been large” would have had you positioned the wrong way round.
The word “tail” here is not the one in our own rulebook
This desk has a standing rule for auction language: a tail means the auction’s high yield came in above the when-issued level, and you do that subtraction yourself before you use the word. We are going to apply that rule against ourselves.
The 0.07 printed here is not that quantity. One of the two readers defines it explicitly as the gap between the average and the lowest accepted prices — a dispersion measure in price terms, describing how far the weakest fill sat from the average fill. It says the bidding was tightly bunched. It says nothing whatsoever about where the auction stopped relative to the pre-auction market.
We have no when-issued level for this sale in this sweep, so we cannot do the subtraction, and we are therefore not going to tell you whether this auction stopped through. We are telling you that 0.07 against 0.15 means the tail narrowed and that is all it means. The same four letters measure two different things in two government bond markets, and the number does not carry its definition with it — which is the third consecutive session in which that has been the most useful thing this desk had to say.
The curve did the work, and the ten-year went the other way
The market reaction, from one reader: the 20-year yield fell as much as 6 basis points to 3.71 per cent, its lowest since 9 September. The 30-year eased 1.5 basis points to 3.970 per cent. The 10-year benchmark rose 1.5 basis points to 2.920 per cent.
So the long end was bought and the ten-year was not. The 20-year to 10-year spread sits at 79 basis points and flattened 7.5 basis points on the day; the 30-year to 20-year segment is 26 basis points. This is a demand event located in one part of the curve, not a rally in Japanese duration.
And here is the cleanest cross-check available this morning, which is worth more than either figure alone: the auction’s average yield of 3.7230 per cent and the post-auction market level of 3.71 per cent sit 1.3 basis points apart. Two sources, measured differently, landing on the same place. When you have no primary document, two independent readings that agree to a basis point are the best evidence you are going to get.
Nine days before the Bank of Japan, the yen declined to care
The policy meeting is 29 and 30 October, with the decision and the Outlook Report’s basic view announced on the 30th and the Governor’s press conference at 06:30 UTC. The policy rate is 1.25 per cent. A record-covered 20-year auction that clears 13 basis points richer is, on its face, the long end telling you it is comfortable funding this government at these yields nine days out.
Dollar-yen did not trade it. Spot came off a 158.55 overnight high to about 158.40 through the Tokyo morning, 15 pips, inside this month’s range and without testing the top. Our refusal here is deliberate and we would rather state it than dress it up: a flattening at the Japanese long end and a yen that goes nowhere do not jointly imply a direction, and we are not giving you one.
What they do imply is a sizing question. If the long end is bid and the ten-year is offered and spot has moved 15 pips, then whatever is going to move dollar-yen this week is not in the JGB auction result, and position size taken on the auction is size taken on a view the market has not expressed. The scheduled risk is elsewhere: UK employment at 06:00 UTC, German and euro area sentiment at 09:00, US housing starts and building permits at 12:30, the New York cut at 14:00.
What this does not tell you
The Ministry of Finance’s own auction result was not read in this sweep. Every figure above is a secondary reading of it, and while the two readers agree on cover and tail, agreement between secondary sources is not the same as a primary document. Treat the metrics as well-attested rather than primary.
The yield levels — 3.71, 3.970 and 2.920 per cent, and the 6 basis point fall — are one reader. The spreads and the flattening derived from them are therefore as single-sourced as the levels they come from, and the 1.3 basis point coherence with the auction average is our arithmetic, not a published reconciliation. The demand attribution, the thirty-year yield high, the record in the 30-year and the insurer’s repatriation comment are all one reader too.
One coincidence we are flagging precisely so nobody builds anything on it: the twelve-month average bid-to-cover is given as 3.73 and this auction’s average yield is 3.7230 per cent. Those are the same digits measuring unrelated quantities in different units. It means nothing.
And one near-miss worth recording. A page surfaced in this morning’s search that looked exactly like what we needed — a 20-year JGB sale, a rally, a tail near a record low — carrying an October display date. Its auction was a different one: cover 3.25, tail 0.009, a 20-year yield of 3.135 per cent and a 10-year at 2.260. Those levels are nowhere near this month’s, and the page itself dates to March. We rejected it on its stamp and the levels confirmed the rejection. Had we taken it, this article would have reported the wrong auction in the right week.
